I do not understand Yellen’s pivot from wages to core inflation

Federal Reserve Chair Janet Yellen gave a speech Friday afternoon in San Francisco that momentarily lit up my Twitter feed:

Lots of the speech rehashed things she’s said before, such as in her recent post-FOMC press conference: the economy is still not back to full speed, but it will likely be time to raise rates later this year, and the path of rate increases will likely be gradual.

But the most interesting part of the speech, as noted by Bloomberg reporter Aki Ito and the New York Times’ Binyamin Appelbaum, was Yellen’s comments on the role of wage growth in the liftoff decision. Appelbaum observed:

In speeches last year, Ms. Yellen highlighted the slow pace of wage inflation as evidence of slack in the labor market. But on Friday she described it as unreliable, saying that wages might be stagnating for a variety of reasons including globalization and the decline of unions.
“The outlook for wages is highly uncertain even if price inflation does move back to 2 percent and labor market conditions continue to improve as projected,” Ms. Yellen said. “This uncertainty limits the usefulness of wage trends as an indicator of the Fed’s progress in achieving its inflation objective.”
Ms. Yellen said that the Fed intended to focus on job growth, judging that if unemployment continued to fall, prices and wages would begin to rise.

This is a pretty striking evolution. Here’s what it looks like in the full context of the speech:

An important factor working to increase my confidence in the inflation outlook will be continued improvement in the labor market. A substantial body of theory, informed by considerable historical evidence, suggests that inflation will eventually begin to rise as resource utilization continues to tighten.2 It is largely for this reason that a significant pickup in incoming readings on core inflation will not be a precondition for me to judge that an initial increase in the federal funds rate would be warranted. With respect to wages, I anticipate that real wage gains for American workers are likely to pick up to a rate more in line with trend labor productivity growth as employment settles in at its maximum sustainable level. We could see nominal wage growth eventually running notably higher than the current roughly 2 percent pace. But the outlook for wages is highly uncertain even if price inflation does move back to 2 percent and labor market conditions continue to improve as projected. For example, we cannot be sure about the future pace of productivity growth; nor can we be sure about other factors, such as global competition, the nature of technological change, and trends in unionization, that may also influence the pace of real wage growth over time. These factors, which are outside of the Federal Reserve’s control, likely explain why real wages have failed to keep pace with productivity growth for at least the past 15 years. For such reasons, we can never be sure what growth rate of nominal wages is consistent with stable consumer price inflation, and this uncertainty limits the usefulness of wage trends as an indicator of the Fed’s progress in achieving its inflation objective.

Yellen says that “considerable historical evidence,” discussed in footnote 2, suggests that the improving labor market will eventually cause higher core inflation, so she doesn’t feel the need to wait for those numbers to come in prior to raising rates. She also doesn’t want to wait for nominal wage growth in the historically expected 3–4% range before raising rates, because a variety of forces outside the Fed’s control are dominating wage-setting dynamics. Since we don’t know what wage growth we should be expecting to get us to 2% core inflation, it doesn’t make sense for the Fed to focus on wage growth in setting rates.

This is the opposite of what I had thought before I read Yellen’s speech. I would have said that we don’t know exactly what the connection between labor market recovery and core inflation will be, but that nominal wage growth is easier to predict based on the position of the labor market. And on re-reading the research Yellen cites (as well as a larger body of research she doesn’t cite), it still seems compatible with the interpretation I had before — a tightening labor market should feature accelerating wage growth.

Yellen’s footnote 2, which presumably includes some of the “considerable historical evidence” she’s looking to, cites four papers:

For recent evidence on the relationship between labor market slack and wages, see Anil Kumar and Pia Orrenius (2014), “A Closer Look at the Phillips Curve Using State Level Data (PDF),” Working Papers 1409 (Dallas: Federal Reserve Bank of Dallas); and Daniel Aaronson and Andrew Jordan (2014), “Understanding the Relationship between Real Wage Growth and Labor Market Conditions,” Chicago Fed Letter No. 327 (Chicago: Federal Reserve Bank of Chicago, October). The price Phillips curve is discussed extensively in the literature; for instance, see Robert J. Gordon (2013), “The Phillips Curve Is Alive and Well: Inflation and the NAIRU during the Slow Recovery,” NBER Working Paper Series 19390 (Cambridge, Mass.: National Bureau of Economic Research, August). In addition, the apparent lack of disinflationary pressure seen during the recent recession is not necessarily a puzzle for the New Keynesian Phillips curve, as shown in Marco Del Negro, Marc P. Giannoni, and Frank Schorfheide (2015), “Inflation in the Great Recession and New Keynesian Models,” American Economic Journal: Macroeconomics, vol. 7 (January), pp. 168–96

The first two papers, by the Dallas Fed and Chicago Fed respectively, both take advantage of variation across states and time to estimate the impacts of labor market conditions (unemployment) on real wage growth. The Dallas Fed paper finds that real wages rise more when the unemployment rate falls, particularly below ~6%:

This paper is what prompted (former!) Dallas Fed President Richard Fisher’s remarks last fall about expecting wage inflation to take off at 6.1% unemployment. (See Tim Duy’s great comments at the time.)

The Chicago Fed paper also finds that real wage growth should tick up as unemployment continues to decline, though it is more focused on which forms of labor market slack best predict real wage growth:

They find that medium-term unemployment and people who are working part time for economic reasons both significantly hamper real wage growth.

Three other papers that Yellen doesn’t cite have recently used labor market variation at the state or metropolitan level to assess the influence of slack on wage growth and consistently found that wage growth should improve as various forms of slack continue to be reduced. (Smith 2014, Blanchflower and Posen 2014, Blanchflower and Levin 2015.)

Given this empirical evidence, which Yellen cites and is very similar to the kinds of evidence typically marshaled in favor of the price Phillips curve she seems to continue to believe in (e.g. Kiley 2014, Fitzgerald and Nicolini 2014), I don’t see what’s driving the move away from wages.

Obviously, I’m not the first person to wonder about this.

Last fall, responding to Fisher’s 6.1% remarks, Tim Duy made a similar observation and posted these two charts:

The first chart shows that as unemployment gets below 6%, wages do seem to start to rise, while the second chart shows the weakness of the connection between wage growth and overall core inflation.

Blanchflower and Posen 2014 says:

Our results also point towards using wage inflation as an additional intermediate target for monetary policy by the FOMC, paralleling on the real activity side the de facto inflation targets on the price stability side. Unemployment has long been known to have severe problems as a guide post to monetary policy, as discussed in Bernanke et al. (1999, chapter 1), although the Phillips curve is far from vertical for extended short runs of multiple years. Guessing the natural rate of unemployment is extremely difficult, is subject to variation, and ignores a lot of additional labor market information. This has all been amply illustrated by the developments of the last few months in the US economy.
By comparison, a general measure of the wage inflation rate encompasses most of the relevant indicators: If mismatch or demographic shifts limit the level of appropriate workers to below the level of demand, wages should be seen to be rising; if on the other hand, individuals are eager for more hours or to return to the workforce, wages should be falling on average for the whole economy.
One has to be careful, as it is possible for wages to rise without generating overall inflation if labor’s share of income rises — and the labor share in the economy is at near-historic lows for the United States. Yet, it is certainly easier and more transparent for the FOMC to assess whether a rebuild in labor share is out of line with historical norms, and/or can be traced to some structural changes in say bargaining power, than to make precise public guesses about the far slipperier NAIRU. And like unemployment, movements in wage growth can be used to predict future movements in inflation. So the FOMC should set its forward guidance for the real economy in terms of wage growth, allowing the economy to recover until wage inflation indicates that labor slack has been absorbed.

Research by Sven Jari Stehn and Jan Hatzius at Goldman Sachs also supports this bottom line, finding that incorporating a focus on wage growth over and above a focus on measured inflation improves welfare in their simulations.

Overall, I think it would be really interesting for someone to ask Yellen what has caused her to update her views on this question. The research she cited, and the rest of what I’ve seen, suggests that we should expect more wage growth as the labor market tightens, and I’m curious what’s led her to move away from wanting to wait for more wage growth before raising rates.

The obvious risk in raising rates before wages rise is that the Fed would tighten before it needs to (because there’s actually more slack than currently estimated), leading to growth below the economy’s long-run potential.